Building infrastructure for self-improving agents @tangle_network
Joined December 2018
- Tweets 3,306
- Following 1,164
- Followers 1,116
- Likes 5,032
141 Photos and videos
We're making an observability system for your @openclaw, @NanoClaw_AI, @NousResearch hermes agents (and more). Anyone wanna be an early user?
It'll give you insights and recommendations into the skills and changes your agent makes to itself and the actions it takes.
1
1
67
If you're interested in building real products with self-improving agents, here's a starter template that comes fully equipped with sandboxes and intelligence to make your product more useful to its users.
github.com/tangle-network/ag…
1
44
Any self-improving agent loop repeatedly looks at scores, modifies itself, and re-evaluates. That is not ordinary benchmarking; it is adaptive reuse of an evaluation set. I'm now able to say I learned something useful, thank you for the Adaptive Data Analysis class @Aaroth 🙏
1
4
712
The new normal is models telling their subagents to do things that trip their safety detectors then shutting you out of superior intelligence. @AnthropicAI losing trust fast.
1
50
There was 0 biology topics in this session or system, therefore it must be cybersecurity or other unmentioned reasons.
The project is a harness for self-improvement.
github.com/tangle-network/ag…
It's crazy, the prompt that led to this is to implement a spec that it itself wrote.
74